As artificial intelligence becomes more integrated into our lives, the data used to train it is under scrutiny. Now, India's IT Ministry has released a handbook for a framework that could change everything about how AI is built here.
What is DEPA?
DEPA stands for the
Data Empowerment and Protection Architecture. It's a framework designed years ago to give Indians control over their personal data, allowing them to share it securely and with explicit consent. Think of it as the next layer of India Stack, which brought us Aadhaar and UPI. While Aadhaar provided identity and UPI enabled payments, DEPA is designed to let individuals leverage their own digital footprint—like financial records or health information—to access services like loans or better medical care, all while maintaining control. It relies on regulated entities called 'Consent Managers' to handle data sharing requests, ensuring you approve every transaction.
Unpacking the New Handbook
The new handbook, released by the Ministry of Electronics and Information Technology (MeitY) in collaboration with the tech think tank iSPIRT Foundation, specifically applies the DEPA framework to the world of AI. It tackles a major problem: how can AI developers in India build powerful, locally relevant models without violating privacy or misusing sensitive data? The solution proposed is radical: instead of data travelling to the AI model, 'the model travels to the data'. This means developers can train their algorithms on valuable datasets held by banks, hospitals, and government agencies without ever seeing or moving the raw data itself. The training happens within secure, controlled digital environments using privacy-preserving techniques.
Data Rights and Consent at the Core
The DEPA for AI Training framework is built on a foundation of user rights and institutional control. It's designed to work in harmony with India's Digital Personal Data Protection Act (DPDP Act), which establishes strict rules around processing personal data. The handbook introduces a techno-legal structure that uses machine-readable contracts and privacy technologies like differential privacy to ensure compliance is a core feature, not an afterthought. This allows data custodians, like hospitals, to let researchers train an AI on medical scans without ever exposing patient records. Furthermore, the framework tracks the origin and use of data throughout the process, ensuring transparency and accountability. This solves a major deadlock where valuable data was locked away due to privacy and liability fears.
Why This Matters for India's AI Ambitions
This new guidance is about more than just rules; it's a strategic move to unlock immense economic value, estimated by MeitY to be around $500 billion for India's GDP by 2030. By creating a trusted mechanism for data sharing, the government aims to prevent a market where only a few large tech giants can afford to build sophisticated AI. This levels the playing field for Indian startups, researchers, and academic labs, giving them access to the high-quality, diverse datasets needed to create solutions for uniquely Indian problems—from extending credit to millions of underserved small businesses to diagnosing diseases prevalent in the local population. The first pilot programs are planned for the crucial finance and healthcare sectors.
The Road Ahead
While groundbreaking, the DEPA for AI framework faces challenges. Its success hinges on the voluntary adoption by data-holding institutions and the robustness of the secure technology it relies on. The handbook has been opened for public recommendations, and the next steps involve launching regulatory sandboxes to test the framework in realworld scenarios. There are also open questions around how exactly consent will be managed for large scale AI training versus individual data use, and how it aligns perfectly with all provisions of the DPDP Act, which comes into full effect in May 2027. However, the framework represents a clear vision for building a sovereign AI capability based on trust, positioning India as a potential leader in ethical AI on the global stage.
